Predictors of retention in an alcohol, tobacco, and other drug prevention study.

This article explored retention patterns, as well as factors that predicted these patterns, in the evaluation of a relationship-based substance abuse prevention intervention study that targeted inner-city African American youth. A total of 851 contacts were made to retain 82% (n = 104) of the baseline sample (N = 127) in the evaluation. Results from multinomial regression analyses indicated that participants who were retained in the evaluation were more likely to perceive alcohol, tobacco, and other drug use as less risky and were more likely to report higher levels of family supervision than were evaluation attrits.